• 한국정보통신학회논문지
• /
• 제25권7호
• /
• pp.962-970
• /
• 2021

대부분의 자동차 사고는 졸음운전과 같은 운전자의 부주의로 인해 발생한다. 전방 추돌 경보 시스템 (FCWS)은 전방 차량으로부터 추돌 위험을 감지하여 운전자에게 사전에 경고함으로써 사고의 위험을 현저하게 줄여준다. 본 논문은 주행 안전을 위한 저전력 임베디드 기반 FCWS를 소개한다. 단일 카메라로부터 전방 차량에 대해 검출, 추적, 거리를 계산하고 현재 차량의 속도 정보를 통해 충돌시간 (TTC)을 계산한다. 또한 저성능 임베디드 시스템에서 실시간으로 동작하기 위해 높고 낮은 수준의 프로그램 최적화 기법을 소개한다. 이 시스템은 임베디드 시스템에서 사전에 취득해둔 주행 영상을 통해서 테스트 하였다. 최적화 기법을 사용한 결과는 이전에 최적화를 하지 않은 프로세스 보다 실행 시간이 약 170배 향상되었다.

• 한국ITS학회 논문지
• /
• 제20권2호
• /
• pp.43-57
• /
• 2021

자율주행의 핵심기술인 첨단 운전자 지원 시스템(Advanced Driver Assistance Systems) 중 대표기술인 전방충돌경보(Forward Collision Warning)를 대상기술로 선정하여, 주행시뮬레이션 실험 기반의 교통사고 예방효과를 추정하였다. 효과척도로 ①인지반응시간(s) ②감속도(m/s2) ③충돌여부(회)로 선정하여, 전방충돌경보 미설치시와 설치시의 변화량 측정하였다. 실험 시나리오는 운전자 전방의 차량의 급정거하는 시나리오(1)과 옆차로에서 차량이 끼어드는 시나리오(2)를 진행하였으며, 주간/야간으로 구분하였다. 분석결과, 전방충돌경보장치를 설치하였을 경우, 인지반응시간(s)이 감소하였으며, 감속도(m/s2)는 감소하였다. 운전자의 위험상황을 빠르게 감지하여 여유로운 감속을 할 수 있게 되었으며, 그에 따른 전방충돌횟수도 감소한 것으로 분석되었다. 향후 운전자의 운전성향을 반영하고 실험 시나리오를 다양화하면, ADAS의 설치효과를 증대시키고 다른 기술의 효과추정에도 활용될 수 있을 것이다.

• 한국ITS학회 논문지
• /
• 제19권6호
• /
• pp.208-221
• /
• 2020

본 연구는 첨단운전자보조시스템(Advanced Driver Assistance System, ADAS)이 빠르게 보급됨에 따라 표본 프로브 차량에 설치된 ADAS로부터 얻은 개별차량의 궤적 데이터와 전방차량과의 차두거리 데이터를 이용하여 연속류의 교통밀도를 추정 및 분석하는 것을 목적으로 한다. 과거 연속류 교통밀도는 주로 차량검지시스템(Vehicle Detection System, VDS)에서 수집되는 교통량, 속도, 점유율 등의 데이터를 가공하여 추정되거나, CCTV등의 영상정보를 활용하여 직접 차량 대수를 계수하여 추정되었다. 이러한 방식은 교통밀도 추정의 공간적 제약이 있고, 교통 혼잡시 추정의 신뢰도가 낮다는 한계를 보였다. 이에 본 연구에서는 선행연구의 한계를 극복하기 위해 ADAS로부터 수집된 개별차량 궤적 데이터와 차두거리 정보를 활용하여 도로의 공간을 검지하고 일반화된 밀도(Generalized Density)방식을 이용하여 시공간적 교통밀도를 추정한다. 이에 따라 ADAS차량의 표본율에 따른 교통밀도 추정의 정확도를 분석한 결과, 30%의 표본율일 경우 교통밀도 참 값과 약 90% 일치하는 것으로 나타났다. 이를 통해 본 연구는 향후 ADAS 및 자율주행차량이 혼재되는 도로 상황에서 신뢰도 높은 교통밀도 추정을 가능하게 하며 효율적인 교통운영관리에 기여할 수 있을 것으로 판단된다.

• 한국안전학회지
• /
• 제33권1호
• /
• pp.148-161
• /
• 2018

A traffic fatality by young people marked average annual decrease of 4.5% since 2011. Meanwhile, a traffic fatality by senior over 65 years old marked average annual increase of 7.9% for the last five years which means that the annual increase of traffic fatality by senior will be a serious problem. This study started questioning that senior drivers over 65 years old did not retain the same causal factor of fatal traffic accidents and thus extensively analyzed a risk of it by age group quantitatively, dividing the senior driver group into the early, middle and latter stages. Depending on the aging level, the risk of traffic fatality showed a wide difference in seven different types of traffic accidents generally, and happened to increase with latter and middle parts of the senior driver more than the early part. Therefore, this study proposes four policy suggestions: 1) The senior driver need to be offered customized driving educations and the improvement of road environment is also recommended. 2) Political assistance is needed to support and guide a safety related technology installation for the new or existing car. 3) Renewal of driving license and an aptitude test(physical examination, cognitive test) for drivers over 75 years old should take in a less than 3 years and an additional road test is needed as occasion demands. 4) Like the United States and Europe, development and extension of customized treatment guidebook for medical teams who examine senior drivers is needed and establishment of education and administration system that a supervisor of driving license renewal can impose safety restriction and American anonymity reporting system is considered to institutionalize in the medium to longer term.

• 인터넷정보학회논문지
• /
• 제24권1호
• /
• pp.39-47
• /
• 2023

The development of autonomous driving and Advanced Driver Assistance System (ADAS) technology has grown rapidly in recent years. As most traffic accidents occur due to human error, self-driving vehicles can drastically reduce the number of accidents and crashes that occur on the roads today. Obviously, technical advancements in autonomous driving can lead to improved public driving safety. However, due to the current limitations in technology and lack of public trust in self-driving cars (and drones), the actual use of Autonomous Vehicles (AVs) is still significantly low. According to prior studies, people's acceptance of an AV is mainly determined by trust. It is proven that people still feel much more comfortable in personalized ADAS, designed with the way people drive. Based on such needs, a new attempt for a customized ADAS considering each driver's driving style is proposed in this paper. Each driver's behavior is divided into two categories: assertive and defensive. In this paper, a novel customized ADAS algorithm with high classification accuracy is designed, which divides each driver based on their driving style. Each driver's driving data is collected and simulated using CARLA, which is an open-source autonomous driving simulator. In addition, Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU) machine learning algorithms are used to optimize the ADAS parameters. The proposed scheme results in a high classification accuracy of time series driving data. Furthermore, among the vast amount of CARLA-based feature data extracted from the drivers, distinguishable driving features are collected selectively using Support Vector Machine (SVM) technology by comparing the amount of influence on the classification of the two categories. Therefore, by extracting distinguishable features and eliminating outliers using SVM, the classification accuracy is significantly improved. Based on this classification, the ADAS sensors can be made more sensitive for the case of assertive drivers, enabling more advanced driving safety support. The proposed technology of this paper is especially important because currently, the state-of-the-art level of autonomous driving is at level 3 (based on the SAE International driving automation standards), which requires advanced functions that can assist drivers using ADAS technology.

• 한국정밀공학회지
• /
• 제21권4호
• /
• pp.88-94
• /
• 2004

Recently, vehicle simulators are widely used to evaluate driver's responses and driver assistance systems. It needs much effort to construct the virtual driving environment for a vehicle simulator. In this study, it is described how to make effectively the roads and the driving environment for a vehicle simulator. The GIS (Geographic Information System) is used to construct the roads and the environment effectively. Because the GIS is the integrated system of geographical data, it contains useful data to make virtual driving environment. First, the outline and centerline of roads is abstracted from the GIS. From the road outline, the road width is calculated. Using the centerline, the grid model of roads is constructed. The final graphic model of roads is constructed by mapping road image to the grid model according to the number of lanes and the kind of surface. Data of buildings from the GIS are abstracted. Each shape and height of buildings is determined according to kind of buildings, the final graphic model of buildings is constructed. Then, the graphic model of roadside tree is also constructed. Finally, the driving environment for driving simulator is constructed by converting the three graphic models with the graphic format of Direct-X and by joining the three graphic models.

• International Journal of Precision Engineering and Manufacturing
• /
• 제6권4호
• /
• pp.3-7
• /
• 2005

Recently, vehicle simulators are widely used to evaluate driver s responses and driver assistance systems. It needs much effort to construct the virtual driving environment for a vehicle simulator. In this study, it is described how to make effectively the roads and the driving environment for a vehicle simulator. GIS (Geographic Information System) is used to construct the roads and the environment effectively. Because the GIS is the integrated system of geographical data, it contains useful data to make virtual driving environment. First, boundaries and centerlines of roads are extracted from the GIS. From boundaries, the road width is calculated. Using centerlines, mesh models of roads are constructed. The final graphic model of roads is constructed by mapping road images to those mesh models considering the number of lanes and the kind of surface. Data of buildings from the GIS are extracted. Each shape and height of building is determined considering the kind of building to construct the final graphic model of buildings. Then, the graphic model of roadside trees is constructed to decide their locations. Finally, the driving environment for driving simulator is constructed by converting the three graphic models with the graphic format of Direct-X and by joining the three graphic models.

• 자동차안전학회지
• /
• 제14권3호
• /
• pp.60-64
• /
• 2022

In the automotive industry, the development of automobiles to meet safety requirements is becoming increasingly complex. This is because quality evaluation agencies in each country are continually strengthening new safety standards for vehicles. Among these various requirements, collision safety must be satisfied by controlling airbags, seat belts, etc., and can be defined as post-crash safety. Apart from this safety system, the Advanced Driver Assistance Systems (ADAS) use advanced detection sensors, GPS, communication, and video equipment to detect the hazard and notify driver before the collision. However, research to improve passenger safety in case of an accident by using the sensor of active safety represented by ADAS in the existing passive safety is limited to the level that utilizes the sudden braking level of the FCA (Forward Collision-avoidance Assist) system. Therefore, this study aims to develop logic that can improve passenger protection in case of an accident by using ADAS information and driving information secured before a collision. The proposed logic was constructed based on LSTM deep learning techniques and trained using crash test data.

• 전기전자학회논문지
• /
• 제20권4호
• /
• pp.351-360
• /
• 2016

최근 딥 러닝을 중심으로 빠르게 발전하고 있는 기계학습 분류 알고리즘은 기존의 방법들보다 뛰어난 성능으로 인하여 주목받고 있다. 딥 러닝 중에서도 Convolutional Neural Network(CNN)는 영상처리에 뛰어나 첨단 운전자 보조 시스템(Advanced Driver Assistance System : ADAS)에서 많이 사용되고 있는 추세이다. 하지만 차량용 임베디드 환경에서 CNN을 소프트웨어로 동작시켰을 때는 각 Layer마다 연산이 반복되는 알고리즘의 특성으로 인해 수행시간이 길어져 실시간 처리가 어렵다. 본 논문에서는 임베디드 환경에서 CNN의 실시간 처리를 위하여 Convolution 연산 및 기타 연산들을 병렬로 처리하여 CNN의 속도를 향상시키는 하드웨어 구조를 제안한다. 제안하는 하드웨어의 성능을 검증하기 위하여 Xilinx ZC706 FPGA 보드를 이용하였다. 입력 영상은 $36{\times}36$ 크기이며, 동작주파수 100MHz에서 하드웨어 수행시간은 약 2.812ms로 실시간 처리가 가능함을 확인했다.

• 한국방송∙미디어공학회:학술대회논문집
• /
• 한국방송공학회 2015년도 하계학술대회
• /
• pp.516-517
• /
• 2015

The automatic traffic sign detection and recognition has been one of the highly researched and an important component of advanced driver assistance systems (ADAS). They are designed especially to warn the drivers of imminent dangers such as sharp curves, under construction zone, etc. This paper presents a traffic sign recognition (TSR) system using morphological operations and multiple descriptors. The TSR system is realized in three stages: segmentation, shape classification and recognition stage. The system is designed to attain maximum accuracy at the segmentation stage with the inclusion of morphological operations and boost the computation time at the shape classification stage using MB-LBP descriptor. The proposed system is tested on the German traffic sign recognition benchmark (GTSRB) and on Korean traffic sign dataset.